Control of cooperative manipulator-endowed systems under high-level tasks and uncertain dynamics

This paper considers the problem of distributed motion- and task-planning multi-agent multi-agent-object systems under temporal-logic-based tasks uncertain dynamics. We focus on manipulator-endowed robotic agents that can interact with their surroundings. present first continuous control algorithms for navigation cooperative object manipulation exhibit following properties. First, they are in sense each agent calculates its own signal from local interaction other environment. Second, guarantee safety properties terms inter-agent collision avoidance obstacle avoidance. Third, adapt on-the-fly to dynamic uncertainties robust exogenous disturbances. The aforementioned allow abstraction underlying system a finite-state representation. Inspired by formal-verification techniques, we use such representation derive plans satisfy given temporal-logic tasks. Various simulation results hardware experiments verify efficiency proposed algorithms.